1. Field of the Invention
The invention pertains to improved electrodialysis ("ED" including "EDR") apparatus and systems including improved filled cell electro-dialysis apparatus and systems and to improved processes which use such apparatus and systems. (Filled cell ED is also known as electrodeionization ("EDI"). Filled cell EDR is also known as reversing electrodeionization ("EDIR")).
2. Description of Prior Art
ED apparatus having a multiplicity of alternating anion selective and cation selective membranes was apparently first described by K. Meyer and W. Strauss in 1940 (Helv. Chim. Acta 23 (1940) 795-800). The membranes used were poorly ion selective. The discovery of ion exchange ("IX") membranes (U.S. Re. No. 24,865) which had high ion perm-selectivity, low electrical resistance and excellent stability led rapidly to the invention of ED using such membranes (U.S. Pat. No. 2,636,852) and to the growth of an industry using such apparatus, for example, for desalting of brackish water, concentration of sea water and deashing of cheese whey. During the last 40 years approximately 5000 ED plants have been installed on a world-wide basis.
Limitations on ED include:
A. Limiting current density. Because the IX membranes used in ED are highly selective to ions of one sign or the other, a substantial fraction of the ions passing through the membranes must reach the latter by diffusion from the ambient solution through laminar flow layers at the interfaces between the membranes and the solutions being depleted of ions (the "dilute or diluting solutions or streams" as known in the art). The maximum rate of diffusion occurs when the concentration of electrolyte at such membrane interfaces is essentially zero. The current density corresponding to such zero concentration is referred to in the art as the limiting current density. To increase the latter it is necessary to increase the rate of diffusion, for example, by reducing the thickness of the laminar flow layers by flowing the ambient solution rapidly by the membrane surfaces and/or by the use of turbulence promoters. Nevertheless practical limiting current densities are generally in the range of 5,000 to 10,000 amperes per square meter for each kilogram-equivalent of salts per cubic meter of solution (that is 0.5 to 1 amperes per square centimeter for each gram-equivalent per liter). A typical brackish water has a concentration of salts of about 0.05 kg-eq/m.sup.3 (that is about 0.05-eq/l or about 3000 parts per million ("ppm")) and therefore a limiting current density in the range of from about 250 to 500 amperes per m.sup.2 (0.025 to 0.05 amperes per cm.sup.2). In order to maximize the utilization of ED apparatus it is desirable to operate at the highest possible current densities. However as the limiting current density is approached it is found that water is dissociated ("split") into hydrogen ions and hydroxide ions at the interfaces between the (conventional) anion exchange ("AX") membranes and the diluting streams. The hydrogen ions pass into the diluting streams and the hydroxide ions into the adjacent solutions which are being enriched in ions (the "concentrate, concentrated, concentrating or brine solutions or streams" as known in the art). Since brackish water may often contain calcium bicarbonate there is a tendency therefore for calcium carbonate to precipitate at those surfaces of the (conventional) AX membranes which are in contact with the concentrating streams. This problem has been addressed by chemical or IX softening of the feed waters or the concentrating streams; by adding acid to the feed waters or the concentrating streams (with or without decarbonation) or by regularly reversing the direction of passage of the electric current thereby changing the concentrating streams to diluting streams (and the diluting streams to concentrating streams). See U.S. Pat. No. 2,863,813. Of the above, the most successful process has been the last mentioned process referred to in the art as "electrodialysis reversal" ("EDR"). PA0 B. Removal of poorly ionized substances:
The theory of limiting current in ED shows that in the case of sodium chloride solution for example, the cation exchange ("CX") membranes should reach their limiting current density at values which are about 2/3rds that of the AX membranes. Careful measurements have shown that such is indeed the case. However as the limiting current density of (conventional) CX membranes is approached or exceeded it is found that water is not split into hydroxide ions and hydrogen ions at the interfaces between such CX membranes and the diluting streams. The difference in behavior of (conventional) AX and CX membranes at their respective limiting currents has been explained in recent years as catalysis of water splitting by weakly basic amines in the AX membranes. AX membranes which have only quaternary ammonium anion exchange groups (and no weakly basic groups) initially do not significantly split water as their limiting current is approached. Such behavior continues for only several hours after which period water splitting begins and increases with time. It is found that the AX membranes then contain some weakly basic groups which have resulted from hydrolysis of quaternary ammonium groups. It is concluded that splitting of water at conventional AX membranes at or near their limiting current densities is an unfortunate phenomenon, unavoidable for practical purposes.
The existence of limiting current also means that in dilute solutions the practical current densities are very low. For example at a concentration of salts of about 0.005 kg-eq/m.sup.3 (that is about 0.005 g-eq/l or about 300 ppm, a concentration typical of drinking water) the limiting current density is in the range of from about 25 to 50 amperes per m.sup.2 (0.0025 to 0.005 amperes per cm.sup.2), i.e. the transfer of salts per unit area per unit time is very low (e.g. 50 to 100 grams of salt per hour per square meter). This problem seems first to have been solved by W. Walters et al. in 1955 (Ind. Eng. Chem. 47 (1955) 61-67) who filled the diluting stream compartments in an ED stack with a mixture of strong base and strong acid IX granules. Since then many patents have issued on the subject, among them U.S. Pat. Nos. 3,149,061; 3,291,713; 4,632,745; 5,026,465; 5,066,375; 5,120,416; and 5,203,976. Two modes of operation of such filled cell ED (that is EDI) have been identified. In the first, the IX granules serve as extensions of the membrane surface area thereby greatly increasing the limiting current density. In the second mode a current density is applied which is very much greater than the limiting current density even with the presence of the IX granules. Under these circumstances the rate of water splitting is very high and the IX granules are predominantly in the strong base and strong acid forms respectively. The apparatus in this mode is therefore best described as continuously electrolytically regenerated (mixed bed) ion exchange. (An intermediate mode may be identified in which there is some water splitting but the IX granules are not predominantly in the strong base and strong acid forms resp.).
Most filled cell ED (that is EDI) operates in both modes e.g. in the same cell (first mode near the entrance to the cell, second mode near the exit); in cells in flow series between a single pair of electrodes; or in separate stacks in flow series (each stack with its own pair of electrodes). Filled cell ED is used to replace conventional, chemically regenerated IX systems e.g. strong acid CX column followed by a weakly basic AX column or, at least in part, a mixed bed IX column. In either of the latter cases the CX and AX granules are chemically regenerated separately e.g. with aqueous solutions of sulfuric acid or hydrochloric acid and sodium hydroxide respectively. Precipitates of calcium carbonate, calcium sulfate and magnesium hydroxide are thereby not obtained. The columns of fine granules are effective filters for colloid matter which is rinsed off the granules during the chemical regeneration. In contrast, in the case of EDI any calcium bicarbonate and/or sulfate removed from the diluting stream occurs in a higher concentration in the concentrating stream, particularly when it is desired to achieve high recoveries of the diluting stream (which is the usual case). Such higher concentrations frequently result in precipitation in the concentrating stream. Furthermore, it is inconvenient (though technically possible) to back-wash the IX granules in a filled cell ED apparatus thereby removing any colloidal matter which may have been filtered out. These problems with EDI are generally solved by pretreatment for example:
regenerable cation exchange for softening followed by regenerable anion exchange absorbents for colloid removal; PA1 ultrafiltration or microfiltration for colloid removal followed by EDR for softening and partial demineralization; or PA1 ultrafiltration or microfiltration for colloid removal followed by nanofiltration for softening or reverse osmosis for softening and partial demineralization.
As pointed out above, filled cell ED is used to replace, at least in part, a mixed bed IX column. The latter however generally produces water having an electrical resistivity of about 18 meg ohm-cm and silica concentrations near the present limits of detection. Such performance by filled cell ED(EDI) has been difficult to achieve until now.
ED (including EDR) is used in many plants to deash cheese whey. Generally the natural whey is first concentrated to the range of 20 to 25 percent solids by weight. The current density (that is the rate of removal of ash per unit area of membrane per unit time) during ED (or EDR) of such concentrated whey remains high until about 50 to 60 percent of the ash is removed. The remaining ash behaves as if it is poorly ionized, perhaps associated or complexed with protein in the whey. An important market for deashed whey requires 90 percent or more deashing. To deash from 40 percent ash to 10 percent ash by ED (including EDR) may require much more apparatus contact time than to deash from 100 percent to 40 percent ash. The problem may be solved by the more or less continuous addition of acid to the whey during deashing from 40 to 10 percent ash, the acid apparently freeing the ash from the protein. However such added acid is rapidly removed by ED (including EDR) and the quantities of acid required are therefore undesirable. The problem has also been solved by removing about 60 percent of the whey ash by ED (including EDR) and removing most of the remaining 40 percent by ion exchange. The latter generally consists of a column of strong acid CX granules followed by a column of weak base AX granules. Considerable quantities of acid and base are required to regenerate the IX granules.